摘要
随着中医药学在国际医疗地位中的不断提高,中草药的使用广泛度也在增加。智能化的中草药图像识别可以进一步推进中医药学的现代化和国际化。传统卷积神经网络的参数量过多,需要大量计算资源,提出一种轻量型卷积神经网络中草药分类算法。新算法使用可分离卷积方式对原始图像与其梯度图的组合进行特征提取,在降低参数量的情况下提高分类的精度。
As Chinese medicine gains a greater importance in global medical treatment,Chinese herbal medicine has been widely used.Therefore,an intelligent image recognition system of Chinese herbal medicine can further facilitate the modernization and internationalization of Chinese medicine.Due to the numerous parameters and computing resources in the traditional convolutional neural network,a lightweight neural network of Chinese herbal medicine classification algorithm is given.The new algorithm is enabled to extract features from the combination of the original image and gradient map,which helps promote the accuracy of classification.
作者
齐保峰
刘华明
王先传
毕学慧
QI Baofeng;LIU Huaming;WANG Xianchuan;BI Xuehui(College of Computer and Information Engineering,Fuyang Normal University,Fuyang 236037,China)
出处
《洛阳理工学院学报（自然科学版）》
2023年第1期73-78,共6页
Journal of Luoyang Institute of Science and Technology：Natural Science Edition
基金
安徽省高校自然科学基金项目(KJ2020ZD46)
阜阳师范大学校级项目(2021FSKJ01ZD)
安徽省社科规划青年项目(AHSKQ2021D47)
阜阳师范大学市校合作科技专项项目(SXHZ202103)
阜阳师范大学校级项目(2020KYQD0032)
阜阳师范大学校级项目(2020FSKJ08ZD).
关键词
神经网络
卷积核
特征组合
通道卷积
neural network
convolution kernel
feature combination
channel convolution
